- Email: [email protected]
MRI-based Preplanning Using CT and MRI Data Fusion in Patients With Cervical Cancer Treated With 3D-based Brachytherapy: Feasibility and Accuracy Study
International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Gynecologic Cancer
MRI-based Preplanning Using CT and MRI Data Fusion in Patients With Cervical Cancer Treated With 3D-based Brachytherapy: Feasibility and Accuracy Study Martin Dolezel, M.D., Ph.D.,*,x Karel Odrazka, M.D., Ph.D.,*,x Jan Zizka, M.D., Ph.D.,{ Jaroslav Vanasek, M.D., Ph.D.,* Tereza Kohlova, M.D.,* Tomas Kroulik, M.Sc.,* Dusan Spitzer, M.D., Ph.D.,y Pavel Ryska, M.D.,y Michal Tichy, M.D.,z Milan Kostal, M.D., Ph.D.,z and Lubica Jalcova, B.Sc.* *Oncology Centre and Departments of yRadiology and zGynaecology, Multiscan and Pardubice Regional Hospital, Pardubice, Czech Republic; xFirst Faculty of Medicine, Charles University, Prague, Czech Republic; and {Department of Radiology, Charles University Teaching Hospital, Hradec Kralove, Czech Republic Received Dec 22, 2010, and in revised form Oct 30, 2011. Accepted for publication Nov 1, 2011
Summary The purpose of this study is to analyze the feasibility and accuracy of computed tomography (CT) and magnetic resonance imaging (MRI) data fusion for MRIbased treatment planning in an institution where an MRI scanner is not available in the radiotherapy department. The registration inaccuracy of applicators and soft tissue was assessed in 42 applications with CT/MRI data fusion. Our results suggest that this technique is feasible and safe with an acceptable
Purpose: Magnetic resonance imaging (MRI)-assisted radiation treatment planning enables enhanced target contouring. The purpose of this study is to analyze the feasibility and accuracy of computed tomography (CT) and MRI data fusion for MRI-based treatment planning in an institution where an MRI scanner is not available in the radiotherapy department. Methods and Materials: The registration inaccuracy of applicators and soft tissue was assessed in 42 applications with CT/MRI data fusion. The absolute positional difference of the center of the applicators was measured in four different planes from the top of the tandem to the cervix. Any inaccuracy of registration of soft tissue in relation to the position of applicators was determined and doseevolume parameters for MRI preplans and for CT/MRI fusion plans with or without target and organs at risk (OAR) adaptation were evaluated. Results: We performed 6,132 measurements in 42 CT/MRI image fusions. Median absolute difference of the center of tandem on CT and MRI was 1.1 mm. Median distance between the center of the right ovoid on CT and MRI was 1.7 and 1.9 mm in the laterolateral and anteroposterior direction, respectively. Corresponding values for the left ovoid were 1.6 and 1.8 mm. Rotation of applicators was 3.1 . Median absolute difference in position of applicators in relation to soft tissue was 1.93, 1.50, 1.05, and 0.84 mm in the respective transverse planes, and 1.17, 1.28, 1.27, and 1.17 mm in selected angular directions. The dosimetric parameters for organs at risk on CT/MRI fusion plans without OAR adaptation were significantly impaired whereas the target coverage was not influenced. Planning without target adaptation led to overdosing of the target volume, especially high-risk clinical target volume d D90 88.2 vs. 83.1 (p < 0.05).
Reprint requests to: Martin Dolezel, M.D., Ph.D., Oncology Centre, Multiscan and Pardubice Regional Hospital, Kyjevska 44, Pardubice 532 03, Czech Republic. Tel: þ420 467 436 410; Fax: þ420 467 436 413; E-mail: [email protected] Presented at the 53rd Annual Meeting of the American Society for Radiology and Oncology (ASTRO), Miami Beach, FL, Oct. 2e6, 2011. Int J Radiation Oncol Biol Phys, Vol. 84, No. 1, pp. 146e152, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.11.003
Conflict of interest: none. AcknowledgmentsdThis work has been facilitated by ESTRO TTG and the Department of Radiotherapy at the Medical University of Vienna/ Vienna General Hospital. Many thanks to Sean Kearns (Dumfries and Galloway Royal Infirmary, Dumfries, UK) for his diligent revision and proofreading of the manuscript.
Volume 84 Number 1 2012
inaccuracy of soft-tissue registration.
MRI-based preplanning with CT/MRI fusion in 3D GYN BRT
147
Conclusions: MRI-based preplanning with consecutive CT/MRI data fusion can be safe and feasible, with an acceptable inaccuracy of soft tissue registration. Ó 2012 Elsevier Inc. Keywords: 3D brachytherapy, Cervical cancer, CT/MRI fusion
Introduction Brachytherapy plays an important role in therapeutic management of patients with cervical cancer. With advances in imaging technology, the use of three-dimensional (3D) treatment planning is becoming more popular in brachytherapy (1, 2). Magnetic resonance imaging (MRI) assisted treatment planning for brachytherapy enables the tailoring of dose distribution to the tumor topography while sparing the organs at risk (OAR) of the patient (3e5). Target volume definitions for MRI, and recommendations for concepts in treatment planning, have been published by the Gynaecological Groupe Europe´en de Curiethe´rapiedEuropean Society for Therapeutic Radiology and Oncology’s Working Group (6, 7). The clinical impact of MRI usage in cervical cancer brachytherapy was confirmed by Po¨tter in 2007 (8). Recently, Dimopoulos demonstrated improvement of local control with image-guided brachytherapy in patients with cervical cancer with increased dose (9). However, MRI is not routinely available in radiotherapy departments and the transport of patients with applicators in situ from remote MRI units to radiotherapy departments may increase the risks of changing the position of the applicators, and may provoke procedure related complications such as bleeding or perforation. Because there was no MRI scanner available in our department, MRI evaluation with applicators in situ was incorporated into our clinical practice by using MRI-based preplanning with consecutive computed tomography (CT)/MRI data fusion. To our knowledge, no systematic integration of MRI-based preplanning with CT/MRI image fusion in patients with cervical cancer has previously been published. This study aims to evaluate the feasibility of applicator insertion under paracervical anesthesia directly within the radiology department, to determine the accuracy of CT and MRI data fusion for MRI-based treatment planning, to specify interfraction changes of target volume and organs at risk, and to evaluate the doseevolume parameters for MRI preplans and for CT/MRI fusion plans with and without target and OAR adaptation.
applicator, Varian Medical Systems, Palo Alto, CA; homemade ovoids). The insertion of applicators was performed on all patients at the radiology department under paracervical anesthesia. The bladder was routinely emptied and then instilled with 50 mL of saline. MRI, with brachytherapy applicators in situ, was performed on a 1.5 Tesla system (Signa Excite HD, GE Healthcare, Milwaukee, WI). T2-weighted axial, coronal, and sagittal fast spin echo sequences (3-mm slice thickness, no gap) were performed for target contouring and fusion with CT images. After MRI examination, the applicators were removed. Target volumes and OAR were contoured according to the Gynaecological Groupe Europe´en de Curiethe´rapiedEuropean Society for Therapeutic Radiology and Oncology Working Group recommendations (6, 7). Because the small aperture of our commercially available applicator made the insertion of plastic saline-filled catheters impossible, we focused on precise delineation of the external contours of our applicator by using gel-absorbed vaginal packing. On T2-weighted images, such packing with absorbed lubricating gel is markedly hyperintense compared to the hypointense applicator. We also used homemade ovoids with an exact geometry specification and a homemade CT/MRI-compatible table. Treatment preplan optimization was made manually using the brachytherapy treatment planning system BrachyVision 8.6. (Varian, Palo Alto, CA). Prescribed dose for high-risk (HR) clinical target volume (CTV) D90 was 6 5 Gy (six fractions in 2 weeks after finishing external beam radiotherapy). Dose volume constraints for OAR were 70 Gy equivalent external beam therapy dose using 2 Gy fractionation EQD2 as a maximum dose in the most exposed tissues of the rectum (rectum D2cc) and sigmoid colon (Sigma D2cc), and 90 Gy EQD2 in 2 cm3 of the bladder (bladder D2cc) (6, 7, 10, 11). No volume constraints were applied for the vagina. The following day, insertion of the applicators was performed with congruent topography as on MRI (identical length of the tandem in the uterus; angle between tandem and ovoids; distance between tandem and ovoids; and distance between ovoids). The bladder was (similarly to MRI) emptied and then instilled with 50 mL of saline. After insertion of the applicators, a CT scan was performed.
Methods and Materials CT/MRI fusion Patients and treatment Patients were treated with either whole pelvis 3D conformal radiotherapy or intensity-modulated external beam radiotherapy using a linear accelerator with 18-MV photons. Concomitant chemotherapy was added using weekly cisplatin 40 mg/m2 of body surface for 5 weeks. Dose was prescribed at the ICRU point and was 45 Gy in 25 fractions (1.8 Gy per fraction). All patients signed a written consent before treatment. The patients underwent MRI scanning before the start of external beam radiotherapy and then again, after finishing external beam radiotherapy before the first and fourth fraction of brachytherapy with applicators in place (GammaMed Fletcher-style
Applicators were used for registration to maximize the accuracy of CT and MRI image fusion within the soft tissue. Primarily, three applicator landmarks were used: the top of the tandem (tandem landmark) and the channels for the interstitial needle located nearest to the rectum (ovoid landmarks). The channels are very well visualized on CT and, because of the gel used, also on MRI. In rare cases of deformed ovoids, or different mutual positions of the tandem and ovoids on CT and MRI, additional landmarks were used. Having registered the match points, the fusion was corrected manually using registration tools of the treatment planning software which enabled shifts and rotation in all planes. All image fusions were performed by the same person.
148
Dolezel et al.
International Journal of Radiation Oncology Biology Physics
3D planning After the registration procedure was completed, we reconstructed the applicators using CT images. The contours of OAR were modified according to current topography on the CT images. Gross tumor volume (GTV), HR CTV, and intermediate-risk (IR) CTV contours were used from the MRI images. Treatment times for dwell positions were transferred from the preplan. Because of the modification of the contours of OAR, adaptation of the treatment plan was always necessary. We followed doseevolume constraints for OAR and for target identical to those used for preplanning.
Registration evaluation The accuracy of CT/MRI registration was evaluated in 6,132 measurements from 42 applications in 7 consecutive patients with CT/MRI data fusion performed in two steps, assessing fusion agreement of both applicators and soft tissue. Every image fusion was monitored by 146 measurements. First, the position of the applicators on CT and MR images was compared using 14 measurements. Absolute positional differences of the center of tandem on CT and MRI was evaluated in four transverse planes (top of the tandem, 15 mm and 30 mm from the top of the tandem, and 10 mm above the external orifice of the cervical canal), in both anteroposterior and laterolateral directions. Differences between the centers of the ovoids on CT and MRI scans were also measured in both directions. Additionally, both the distance between the center of the ovoids and applicator rotation were evaluated. Second, an analysis of registration accuracy of soft tissue was performed. Because of the unequal volume of the cervix on CT and MRI images, differences in the relative position of the applicator within the uterine cervix and body were determined on CT and MRI scans using the ratio of the distance between the applicator and external uterine contour to the uterine diameter in eight angular projections (0 e180 , 45 e225 , 90 e270 , 135 e315 ) and four transverse planes (top of the tandem, 15 mm and 30 mm from the top of the tandem, and 10 mm above the external orifice of the cervical canal). The difference in the relative applicator position on CT and MRI was converted to the absolute distance in millimeters which indicated the difference between the real and ideal applicator position (i.e., agreement of soft-tissue fusion) (Fig. 1aeb). Measurements were performed manually using TPS tools.
Dosimetric evaluation Four plans were compared for dosimetric evaluation. Besides the two plans used for treatment planning (MRI preplan and definitive CT/MRI fusion plan), other two modifications of CT/MRI fusion plans were created to analyze the dosimetric impact of volume variability: CT/MRI fusion plan without OAR adaptation and CT/ MRI fusion plan without target adaptation. CT/MRI fusion plan without OAR adaptation was created after CT/MRI fusion by using dwell times from MRI preplan regardless of the changed topography of organs at risk (rectum, sigmoid, bladder) to assess the need for inter-fraction reimaging. CT/MRI fusion plan without target adaptation was planned by using target volume contours from MRI performed before first brachytherapy fraction; however, doseevolume histogram
Fig. 1. (a) Diagram of fusion analysis (soft tissue). The ratio of the distance between the applicator and uterine margin to uterine diameter was determined on computed tomography (CT) and magnetic resonance imaging (MRI) scans in eight directions, at angular intervals of 45 . Accuracy of registration of soft tissue was evaluated for each fraction on transverse images in 15-mm steps from the top of the tandem to the cervix, using at least 132 measurements. (Example of absolute difference of the relative position of the applicator in direction 90 e270 Z a/b - d/c.) (b) Diagram of fusion accuracy assessment (soft tissue). The absolute difference of the relative position was converted to the absolute distance in millimeters (z). In this example, in direction 90 e270 . parameters for target volumes (GTV, HR CTV, IR CTV) were reported to MRI performed before fourth brachytherapy fraction to determine the dosimetric impact of spatial variability of target volume contours. In addition, the GTV volume regression during radiotherapy was measured (before start of external beam radiotherapy, before fraction 1 of brachytherapy, and before fraction 4 of brachytherapy).
Statistics All continuous numerical variables were presented as mean values with ranges and standard deviations. Differences in position of the
Volume 84 Number 1 2012 Table 1
MRI-based preplanning with CT/MRI fusion in 3D GYN BRT
149
Inaccuracy of CT/MRI applicator registration Laterolateral direction (mm) (mean [SD]) (median [range])
Tandem 1.10 Top of the tandem 1.28 15 mm below top of the tandem 1.16 30 mm below top of the tandem 1.06 10 mm above the external orifice of cervical canal 1.14 Ovoid Center of the left ovoid 1.68 Center of the right ovoid 1.73 Distance between center of the ovoids 2.88 mm Rotation of the applicator 3.32
(0.61) (0.83) (0.52) (0.45) (0.42)
1.10 1.15 1.15 1.05 1.00
(0.30e3.60) (0.60e3.60)* (0.40e2.60)* (0.30e2.60)* (0.3e2.40)*
(0.67) (0.73) (0.86) (1.56)
1.60 (0.50e4.80) 1.70 (0.40e5.10)
Anteroposterior direction (mm) (mean [SD]) (median [range]) 1.20 1.52 1.38 1.08 0.98 1.87 1.96 2.85 mm 3.10
(0.68) (0.82) (0.56) (0.41) (0.34)
1.10 1.50 1.25 1.10 1.00
(0.20e3.40) (0.40e3.40)y (0.30e2.90)y (0.20e2.10)y (0.30e1.90)y
(0.72) 1.80 (0.50e3.60) (0.65) 1.90 (0.40e3.80) (0.80e5.10) (1.00e7.40)
Abbreviation: SD Z standard deviation. * The differences between measured values were not significant. y The differences between measured values were significant between planes with a distance of 30 mm or more (top of the tandem vs. 30 mm below top of the tandem or 10 mm above the external orifice of cervical canal, and 15 mm below top of the tandem vs. 10 mm above the external orifice of cervical canal).
applicators and doseevolume histogram parameters were assessed by t-test. A p value of 0.05 or less was considered significant.
Results The insertion of applicators under paracervical anesthesia in the radiology department was successful in all patients. We observed no complications during the entire MRI procedure at the radiology department (insertion; MRI scanning; applicator removal and subsequent transport). The average time of insertion of applicators in the MRI room under paracervical anesthesia was 15 min. The average contouring and planning time needed for preplanning was 77 min (range, 55e94 min). The average time needed for CT/MRI data fusion, modification of OAR contours, and plan adaptation was 34 min (range, 25e49 min). When assessing the accuracy of applicator registration, the median absolute difference of the center of the tandem on CT and MRI was 1.1 mm in both the anteroposterior and laterolateral directions. There was no statistical difference in the laterolateral direction. However, a statistical difference in the anteroposterior direction between planes of 30 mm or more distance from each other was observed. In more than 85% of measurements, the absolute difference of the tandem on CT and MRI was less than 1.5 mm. In only 7.5% of measurements, the difference was greater than 2 mm. The median distance between the center of the right ovoid on CT and MRI was 1.7 mm and 1.9 mm in the laterolateral and anteroposterior directions, respectively. Similarly, the median distance between the center of the left ovoid was 1.6 mm and 1.8 mm, respectively. More than 93% of measurements had a difference in the distance between right or left ovoid smaller than 2.5 mm. The median absolute difference between the separation of the center of ovoids on CT and MRI was 2.85 mm. The rotation of applicators was 3.1 (Table 1). In terms of soft-tissue registration accuracy, the median absolute difference in the position of the applicator in relation to soft tissue was reduced from 1.93 mm in the transverse plane passing through the top of the tandem to 0.84 mm in the plane 10 mm above the external orifice of the cervical canal. The differences were statistically significant. Similarly, the median absolute difference in the applicator position in relation to soft tissue was
between 1.17 mm and 1.28 mm in selected directions (angles 0 e180 , 45 e225 , 90 e270 , 135 e315 ), without statistical significance (Table 2, Fig. 2). In 80% of measurements, the difference was smaller than 2 mm. An inaccuracy greater than 3 mm was found in only 7.1% of measurements. The degree of inaccuracy of applicator/applicator fusion correlated with the inaccuracy of soft-tissue registration. Three of 42 evaluated fusions were found unacceptable, with an absolute difference of the tandem on CT and MRI of more than 2.0 mm, an absolute difference of ovoid on CT and MRI of more than 2.5 mm, and an inaccuracy of soft tissue greater than 3.0 mm. All incorrect fusions were observed in a single patient. The mean volumes of GTV before treatment, before fraction 1 of brachytherapy, and before fraction 4 of brachytherapy were 65.4 cm3, 12.6 cm3, and 7.8 cm3, respectively. The mean dose
Table 2
Inaccuracy of CT/MRI soft-tissue registration Absolute difference in position of the applicator in relation to soft tissue (mm)
Plane Top of the tandem 15 mm below top of the tandem 30 mm below top of the tandem 10 mm above the external orifice of cervical canal Angle 0 and 180 45 and 225 90 and 270 135 and 315
mean (SD)
median (range)
2.05 (1.48) 1.58 (0.89)
1.93 (0.11e5.87)y 1.50 (0.14e4.23)y
1.12 (0.85)
1.05 (0.13e4.15)y
0.90 (0.64)
0.84 (0.06e3.41)y
1.25 1.37 1.34 1.23
1.17 1.28 1.27 1.17
(1.14) (1.26) (0.88) (0.93)
(0.06e5.64)* (0.11e5.47)* (0.08e4.33)* (0.08e5.87)*
Abbreviation: SD Z standard deviation. * The differences between measured values were not significant. y The differences between measured values were significant.
150
Dolezel et al.
Fig. 2. Distribution of computed tomography/magnetic resonance imaging registration inaccuracy. Each point represents the difference in the position of the applicator in relation to soft tissue: 42 fractions, 4 planes. > Z top of the tandem; O Z 15 mm below top of the tandem; , Z 30 mm below top of the tandem; Δ Z 10 mm above the external orifice of cervical canal.
values to GTV, HR CTV, IR CTV, urinary bladder, sigmoid, and rectum for all four plans are summarized in Table 3. No statistically significant difference in dosimetric parameters between CT/ MRI plans in comparison to MRI preplans was observed; however, dwell times varied significantly. The dosimetric parameters for organs at risk on CT/MRI fusion plans without OAR adaptation were significantly worsened whereas the target coverage was not changed. Planning without target adaptation led to overdosing of the target volume, especially HR CTV.
Discussion Conventional dose calculation for intracavitary brachytherapy was for decades based on the information obtained from two orthogonal radiographs with limited visualization of the organs at risk, and with no information about the tumor itself. Although CT imaging is sufficient for definition of OAR (rectum, bladder, sigmoid), it does not provide sufficient information about the tumor topography nor the distinction between the uterine cervix and corpus. Recently, 3D MRI-guided brachytherapy has been established in clinical practice (2e5, 10, 12e20). The first published clinical results demonstrated a reduction in morbidity and improved local control (13, 14). Because of the absence of an MRI scanner in our radiotherapy department, we aimed to develop an alternative technique of MRI evaluation with applicators in situ using MRI-based preplanning with consecutive CT/MRI data fusion. The accuracy of CT/MRI fusion based on the applicators is highly dependent on the congruent geometry of applicators both at MRI preplanning and CT application. Therefore, we strictly adhered to an identical topography of applicators. We also developed homemade CT/MRI ovoids, compatible table, and homemade MRI-compatible clamping device. As a result, fusion of CT and MRI images with only a 1.1 mm absolute difference in the position
International Journal of Radiation Oncology Biology Physics of tandem and 1.7, respectively 1.8 mm in the position of ovoids has been achieved. Determination of the accuracy of soft-tissue registration is a “sophisticated approximation” because of the unequal volumes of the uterine cervix and body on CT and MRI images. In our opinion, conversion of the absolute difference of the relative position of the applicator in relation to the outer contour of the uterine cervix or body is rather controversial but still the most precise way of evaluation. To maximize the accuracy of our method, we performed a large number of measurements for every CT/MRI fusion (146 measurements per CT/MRI registration). The absolute difference in the position of the applicator in relation to the soft tissue was decreasing towards the cervix (1.93 mm, 1.50 mm, 1.05 mm, 0.84 mm), the differences reached statistical significance. To the contrary, there was no statistical difference in any of the eight measured angular projections. The results are in keeping with our presumption. Because of decreased fixation of the upper part of the uterus in comparison to the cervix, soft-tissue registration based on the applicators showed increasing accuracy toward the cervix. Because the tumor is located in an area with the most precise fusion, we can postulate that this approach is acceptable for image registration. We found unacceptable results of image fusion in three out of 42 evaluated fusions. Interestingly enough, all inappropriate fusions were obtained in the same patient. We observed, in this particular case, significant topographic changes (narrowing) in the upper part of vagina after the first fraction of brachytherapy and a small but significant difference in rotation of the tandem on MRI before the first fraction of brachytherapy compared to the CT scan performed during the first three applications. Because applicators and critical organs are reconstructed and contoured directly on the CT scan, the only, yet very important inaccuracy, applies to the target volumes (GTV, HR CTV, and IR CTV). This fusion inaccuracy of soft tissues must be taken into account during the planning process. We found significant GTV volume regression during external beam chemoradiotherapy and only slight changes of tumor volume during brachytherapy. Treatment times for individual dwell positions differed significantly between preplans and all CT/MRI plans mostly because of the different interfractional topography of OAR. The most variable organ was sigmoid. Minor changes were found for rectum and (likely from routine instillation of 50 mL saline) minimal changes for bladder topography. Therefore, we found the greatest interfractional difference in particular dwell position at the upper part of the tandem (39%e287%), lesser in ovoids (59%e 195%), and minimal in the lower part of the tandem (74%e137%). Due to the substantial interfractional OAR volume changes in comparison with GTV regression during brachytherapy, no significant correlation between the need for OAR adaptation and tumor regression was found. Our dosimetric analysis confirmed the necessity of preplan adaptation and the need for interfraction imaging. Our results show that CT acquisition for every fraction from OAR variability is essential and, in addition, at least two to three MRI acquisitions are advisable because of systematic shrinkage of the HR CTV (Fig. 3, Table 3). The main advantages of MRI-based preplanning with satisfactory CT/MRI data fusion are: precise definition of GTV and HR CTV; increased conspicuity of applicators on CT scans with minimal risk of inaccuracy of applicator reconstruction; decreased time needed for the planning process with a reduction of volume changes of OAR; time-unlimited preplanning (without personal stress and decreased comfort to the patient because of prolonged
Volume 84 Number 1 2012 Table 3
MRI-based preplanning with CT/MRI fusion in 3D GYN BRT
151
Difference in dosimetry between MRI preplan and CT/MRI fusion plans
GTV D100 (Gy a/b10) D90 (Gy a/b10) HR-CTV D100 (Gy a/b10) D90 (Gy a/b10) IR-CTV D100 (Gy a/b10) D90 (Gy a/b10) Bladder ICRU point D0.1cc (Gy a/b3) D1cc (Gy a/b3) D2cc (Gy a/b3) Rectum ICRU point D0.1cc (Gy a/b3) D1cc (Gy a/b3) D2cc (Gy a/b3) Sigmoid D0.1cc (Gy a/b3) D1cc (Gy a/b3) D2cc (Gy a/b3)
MRI preplan (mean SD)
CT/MRI fusion plan without OAR adaptation (mean SD)
CT/MRI fusion plan (mean SD)
CT/MRI fusion plan without target adaptation (mean SD)
93.9 21.2 122.1 28.4
93.7 22.0; *p Z 0.49 122.3 28.6; *p Z 0.61
93.1 22.0; zp Z 0.62 121.3 29.6; zp Z 0.68
95.9 21.3; yp Z 0.10 123.4 29.6; yp Z 0.08
66.1 7.8 83.8 12.7
66.3 7.7; *p Z 0.54 83.9 12.9; *p Z 0.47
65.4 7.6; zp Z 0.59 83.1 13.8; zp Z 0.51
68.5 8.3; yp < 0.05 88.2 11.6; yp < 0.05
53.1 7.4 64.2 7.3
53.2 7.5; *p Z 0.42 64.2 7.3; *p Z 0.39
52.6 7.7; zp Z 0.47 63.8 6.9; zp Z 0.37
54.1 8.2; yp Z 0.13 65.2 7.6; yp Z 0.09
65.5 96.2 82.8 77.5
7.9 6.4 6.1 5.1
66.9 103.4 88.2 84.1
8.8; 9.7; 7.5; 7.3;
*p *p *p *p
Z 0.21 < 0.05 < 0.05 < 0.05
65.7 97.1 83.6 78.2
8.1; 6.7; 6.5; 5.8;
67.2 73.8 67.1 63.6
2.9 3.8 2.7 2.4
72.9 82.3 71.4 69.3
3.4; 3.9; 3.1; 2.4;
*p *p *p *p
< < < <
67.9 74.9 67.5 64.2
2.7; 3.6; 2.3; 1.9;
76.8 4.1 68.7 3.1 64.2 2.7
0.05 0.05 0.05 0.05
90.9 4.5; *p < 0.05 78.7 3.6; *p < 0.05 74.2 3.7; *p < 0.05
z
p p z p z p
Z Z Z Z
0.82 0.54 0.31 0.46
66.1 97.3 84.2 78.6
8.5; 6.9; 6.7; 6.1;
z
Z Z Z Z
0.73 0.62 0.59 0.53
68.3 75.4 67.9 64.6
3.1; 3.9; 2.4; 2.1;
z
p p z p z p z
77.5 3.9; zp Z 0.63 69.4 2.8; zp Z 0.41 65.5 2.9 zp Z 0.38
y
p p y p y p
Z Z Z Z
0.43 0.76 0.31 0.49
y
Z Z Z Z
0.87 0.74 0.79 0.67
y
p p y p y p y
77.8 4.2; yp Z 0.92 69.7 3.0; yp Z 0.75 65.9 3.2; yp Z 0.58
Abbreviations: CT Z computed tomography; D90 Z dose received by at least 90% of volume; D0.1cc, D1cc, D2cc Z minimal dose for most irradiated 0.1 cm3, 1 cm3, and 2 cm3; D100 Z minimal target volume; GTV Z gross tumor volume; HR-CTV Z high-risk clinical target volume; ICRU Z International Commission on Radiation Units and Measurements; IR-CTV Z intermediate-risk clinical target volume; MRI Z magnetic resonance imaging; OAR Z organs at risk; SD Z standard deviation. * Comparison of dosimetric parameters between CT/MRI fusion plans without OAR adaptation and CT/MRI fusion plans (t-test). y Comparison of dosimetric parameters between CT/MRI fusion plans without target adaptation and CT/MRI fusion plans (t-test). z Comparison of dosimetric parameters between CT/MRI fusion plans and MRI preplans (t-test).
contouring and planning process); and the possibility of combined interstitial and intracavitary brachytherapy in patients with a suboptimal response to the external chemoradiotherapy. The disadvantages, obviously, are the additional insertion of applicators at the radiology department under paracervical anesthesia before the first and fourth fraction of brachytherapy without radiotherapy treatment, and the necessity of congruent alignment
of applicators on MRI and CT scans. Also, it would be more appropriate to use a deformable algorithm for accurate coregistration of MRI onto CT.
Conclusions This is the first evidence of the systematic integration of MRI based preplanning with consecutive CT/MRI data fusion into clinical practice. Our results suggest that this technique is feasible and safe with an acceptable inaccuracy of soft-tissue registration. Fusion inaccuracy must be taken into account during the planning process.
References
Fig. 3. Mean absolute tumor volumes during radiotherapy. EBRT Z External beam radiotherapy; BRTZ brachytherapy.
1. Po¨tter R, Dimopoulos J, Bachtiary B, et al. 3D-conformal HDR brachytherapy and external beam therapy plus simultaneous cisplatin for high-risk cervical cancer: Clinical experience with 3 year follow up. Radiother Oncol 2006;79:80e86. 2. Po¨tter R, Fidarova E, Kirisits C, et al. Image-guided adaptive brachytherapy for cervix carcinoma. Clin Oncol 2008;20:426e432. 3. Dimopoulos JCA, Schard G, Berger D, et al. Systematic evaluation of MRI findings in different stages of treatment of cervical cancer: Potential of MRI on delineation of target, pathoanatomic structures, and organs at risk. Int J Radiat Oncol Biol Phys 2006;64:1380e1388.
152
Dolezel et al.
4. Dimopoulos JC, Schirl G, Baldinger A, et al. MRI assessment of cervical cancer for adaptive radiotherapy. Strahlenther Onkol 2009; 185:282e287. 5. Haie-Meder C, Chargari C, Rey A, et al. DVH parameters and outcome for patients with early-stage cervical cancer treated with preoperative MRI-based low dose rate brachytherapy followed by surgery. Radiother Oncol 2009;93:316e321. 6. Haie-Meder C, Po¨tter R, van Limbergen E, et al. Recommendations from the Gynaecological (GYN) GEC ESTRO Working Group: Concepts and terms in 3D-image based 3D-treatment planning in cervix cancer brachytherapy with emphasis on MRI assessment of GTV and CTV. Radiother Oncol 2004;74:235e245. 7. Po¨tter R, Haie-Meder C, vanLimbergen E, et al. Recommendations from gynaecological (GYN) GEC ESTRO working group (II): Concepts and terms in 3D image-based treatment planning in cervix cancer brachytherapy-3D dose-volume parameters and aspects of 3D-image-based anatomy, radiation physics, radiobiology. Radiother Oncol 2006;78:67e77. 8. Po¨tter R, Dimopoulos J, Georg P, et al. Clinical impact of MRI assisted dose volume adaptation and dose escalation in brachytherapy of locally advanced cervix cancer. Radiother Oncol 2007;83:148e155. 9. Dimopoulos JCA, Lang S, Kirisits C, et al. Doseevolume histogram parameters and local tumor control in MR image guided cervical cancer brachytherapy. Int J Radiat Oncol Biol Phys 2009;75:56e63. 10. Kirisits C, Po¨tter R, Lang S, et al. Dose and volume parameters for MRI based treatment planning in intracavitary brachytherapy of cervix cancer. Int J Radiat Oncol Biol Phys 2005;62:901e911. 11. Lang S, Kirisits C, Dimopoulos J, et al. Treatment planning for MRI assisted brachytherapy of gynaecological malignancies based on total dose constraints. Int J Radiat Oncol Biol Phys 2007;69:619e627. 12. Haack S, Nielsen SK, Lindegaard JC, et al. Applicator reconstruction in MRI 3D image-based dose planning of brachytherapy for cervical cancer. Radiother Oncol 2009;91:187e193.
International Journal of Radiation Oncology Biology Physics 13. Tanderup K, Hellebust TP, Lang S, et al. Consequences of random and systematic reconstruction uncertainties in 3D image based brachytherapy in cervical cancer. Radiother Oncol 2008;89: 156e163. 14. Muschitz S, Petrow P, Briot E, et al. Correlation between the treated volume, the GTV and the CTV at the time of brachytherapy and the histopathologic findings in 33 patients with operable cervix carcinoma. Radiother Oncol 2004;73:187e194. 15. Kirisits C, Lang S, Dimopoulos J, et al. The Vienna applicator for combined intracavitary and interstitial brachytherapy of cervical cancer: Design, application, treatment planning and dosimetric results. Int J Radiat Oncol Biol Phys 2006;65:624e630. 16. Dimopoulos JCA, Kirisits C, Petric P, et al. The Vienna applicator for combined intracavitary and interstitial brachytherapy of cervical cancer: Aspects of clinical feasibility and preliminary results. Int J Radiat Oncol Biol Phys 2006;66:83e90. 17. Van de Kamer JB, De Leeuw AA, Moerland MA, et al. Determining DVH parameters for combined external beam and brachytherapy treatment: 3D biological dose adding for patients with cervical cancer. Radiother Oncol 2010;94:248e253. 18. Ju¨rgenliemk-Schulz IM, Lang S, Tanderup K, et al. Variation of treatment planning parameters (D90 HR-CTV, D 2cc for OAR) for cervical cancer tandem ring brachytherapy in a multicentre setting: Comparison of standard planning and 3D image guided optimisation based on a joint protocol for dose-volume constraints. Radiother Oncol 2010;94:339e345. 19. Beriwal S, Kim H, Coon D, et al. 3D HDR intracavitary brachytherapy for the treatment of cervical cancer: Feasibility and early results from single institution. Int J Radiat Oncol Biol Phys 2009; 75:367e368. 20. Holloway CL, Racine ML, Cormack RA, et al. Sigmoid dose using 3D imaging in cervical-cancer brachytherapy. Radiother Oncol 2009;93: 307e310.
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Gynecologic Cancer
MRI-based Preplanning Using CT and MRI Data Fusion in Patients With Cervical Cancer Treated With 3D-based Brachytherapy: Feasibility and Accuracy Study Martin Dolezel, M.D., Ph.D.,*,x Karel Odrazka, M.D., Ph.D.,*,x Jan Zizka, M.D., Ph.D.,{ Jaroslav Vanasek, M.D., Ph.D.,* Tereza Kohlova, M.D.,* Tomas Kroulik, M.Sc.,* Dusan Spitzer, M.D., Ph.D.,y Pavel Ryska, M.D.,y Michal Tichy, M.D.,z Milan Kostal, M.D., Ph.D.,z and Lubica Jalcova, B.Sc.* *Oncology Centre and Departments of yRadiology and zGynaecology, Multiscan and Pardubice Regional Hospital, Pardubice, Czech Republic; xFirst Faculty of Medicine, Charles University, Prague, Czech Republic; and {Department of Radiology, Charles University Teaching Hospital, Hradec Kralove, Czech Republic Received Dec 22, 2010, and in revised form Oct 30, 2011. Accepted for publication Nov 1, 2011
Summary The purpose of this study is to analyze the feasibility and accuracy of computed tomography (CT) and magnetic resonance imaging (MRI) data fusion for MRIbased treatment planning in an institution where an MRI scanner is not available in the radiotherapy department. The registration inaccuracy of applicators and soft tissue was assessed in 42 applications with CT/MRI data fusion. Our results suggest that this technique is feasible and safe with an acceptable
Purpose: Magnetic resonance imaging (MRI)-assisted radiation treatment planning enables enhanced target contouring. The purpose of this study is to analyze the feasibility and accuracy of computed tomography (CT) and MRI data fusion for MRI-based treatment planning in an institution where an MRI scanner is not available in the radiotherapy department. Methods and Materials: The registration inaccuracy of applicators and soft tissue was assessed in 42 applications with CT/MRI data fusion. The absolute positional difference of the center of the applicators was measured in four different planes from the top of the tandem to the cervix. Any inaccuracy of registration of soft tissue in relation to the position of applicators was determined and doseevolume parameters for MRI preplans and for CT/MRI fusion plans with or without target and organs at risk (OAR) adaptation were evaluated. Results: We performed 6,132 measurements in 42 CT/MRI image fusions. Median absolute difference of the center of tandem on CT and MRI was 1.1 mm. Median distance between the center of the right ovoid on CT and MRI was 1.7 and 1.9 mm in the laterolateral and anteroposterior direction, respectively. Corresponding values for the left ovoid were 1.6 and 1.8 mm. Rotation of applicators was 3.1 . Median absolute difference in position of applicators in relation to soft tissue was 1.93, 1.50, 1.05, and 0.84 mm in the respective transverse planes, and 1.17, 1.28, 1.27, and 1.17 mm in selected angular directions. The dosimetric parameters for organs at risk on CT/MRI fusion plans without OAR adaptation were significantly impaired whereas the target coverage was not influenced. Planning without target adaptation led to overdosing of the target volume, especially high-risk clinical target volume d D90 88.2 vs. 83.1 (p < 0.05).
Reprint requests to: Martin Dolezel, M.D., Ph.D., Oncology Centre, Multiscan and Pardubice Regional Hospital, Kyjevska 44, Pardubice 532 03, Czech Republic. Tel: þ420 467 436 410; Fax: þ420 467 436 413; E-mail: [email protected] Presented at the 53rd Annual Meeting of the American Society for Radiology and Oncology (ASTRO), Miami Beach, FL, Oct. 2e6, 2011. Int J Radiation Oncol Biol Phys, Vol. 84, No. 1, pp. 146e152, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.11.003
Conflict of interest: none. AcknowledgmentsdThis work has been facilitated by ESTRO TTG and the Department of Radiotherapy at the Medical University of Vienna/ Vienna General Hospital. Many thanks to Sean Kearns (Dumfries and Galloway Royal Infirmary, Dumfries, UK) for his diligent revision and proofreading of the manuscript.
Volume 84 Number 1 2012
inaccuracy of soft-tissue registration.
MRI-based preplanning with CT/MRI fusion in 3D GYN BRT
147
Conclusions: MRI-based preplanning with consecutive CT/MRI data fusion can be safe and feasible, with an acceptable inaccuracy of soft tissue registration. Ó 2012 Elsevier Inc. Keywords: 3D brachytherapy, Cervical cancer, CT/MRI fusion
Introduction Brachytherapy plays an important role in therapeutic management of patients with cervical cancer. With advances in imaging technology, the use of three-dimensional (3D) treatment planning is becoming more popular in brachytherapy (1, 2). Magnetic resonance imaging (MRI) assisted treatment planning for brachytherapy enables the tailoring of dose distribution to the tumor topography while sparing the organs at risk (OAR) of the patient (3e5). Target volume definitions for MRI, and recommendations for concepts in treatment planning, have been published by the Gynaecological Groupe Europe´en de Curiethe´rapiedEuropean Society for Therapeutic Radiology and Oncology’s Working Group (6, 7). The clinical impact of MRI usage in cervical cancer brachytherapy was confirmed by Po¨tter in 2007 (8). Recently, Dimopoulos demonstrated improvement of local control with image-guided brachytherapy in patients with cervical cancer with increased dose (9). However, MRI is not routinely available in radiotherapy departments and the transport of patients with applicators in situ from remote MRI units to radiotherapy departments may increase the risks of changing the position of the applicators, and may provoke procedure related complications such as bleeding or perforation. Because there was no MRI scanner available in our department, MRI evaluation with applicators in situ was incorporated into our clinical practice by using MRI-based preplanning with consecutive computed tomography (CT)/MRI data fusion. To our knowledge, no systematic integration of MRI-based preplanning with CT/MRI image fusion in patients with cervical cancer has previously been published. This study aims to evaluate the feasibility of applicator insertion under paracervical anesthesia directly within the radiology department, to determine the accuracy of CT and MRI data fusion for MRI-based treatment planning, to specify interfraction changes of target volume and organs at risk, and to evaluate the doseevolume parameters for MRI preplans and for CT/MRI fusion plans with and without target and OAR adaptation.
applicator, Varian Medical Systems, Palo Alto, CA; homemade ovoids). The insertion of applicators was performed on all patients at the radiology department under paracervical anesthesia. The bladder was routinely emptied and then instilled with 50 mL of saline. MRI, with brachytherapy applicators in situ, was performed on a 1.5 Tesla system (Signa Excite HD, GE Healthcare, Milwaukee, WI). T2-weighted axial, coronal, and sagittal fast spin echo sequences (3-mm slice thickness, no gap) were performed for target contouring and fusion with CT images. After MRI examination, the applicators were removed. Target volumes and OAR were contoured according to the Gynaecological Groupe Europe´en de Curiethe´rapiedEuropean Society for Therapeutic Radiology and Oncology Working Group recommendations (6, 7). Because the small aperture of our commercially available applicator made the insertion of plastic saline-filled catheters impossible, we focused on precise delineation of the external contours of our applicator by using gel-absorbed vaginal packing. On T2-weighted images, such packing with absorbed lubricating gel is markedly hyperintense compared to the hypointense applicator. We also used homemade ovoids with an exact geometry specification and a homemade CT/MRI-compatible table. Treatment preplan optimization was made manually using the brachytherapy treatment planning system BrachyVision 8.6. (Varian, Palo Alto, CA). Prescribed dose for high-risk (HR) clinical target volume (CTV) D90 was 6 5 Gy (six fractions in 2 weeks after finishing external beam radiotherapy). Dose volume constraints for OAR were 70 Gy equivalent external beam therapy dose using 2 Gy fractionation EQD2 as a maximum dose in the most exposed tissues of the rectum (rectum D2cc) and sigmoid colon (Sigma D2cc), and 90 Gy EQD2 in 2 cm3 of the bladder (bladder D2cc) (6, 7, 10, 11). No volume constraints were applied for the vagina. The following day, insertion of the applicators was performed with congruent topography as on MRI (identical length of the tandem in the uterus; angle between tandem and ovoids; distance between tandem and ovoids; and distance between ovoids). The bladder was (similarly to MRI) emptied and then instilled with 50 mL of saline. After insertion of the applicators, a CT scan was performed.
Methods and Materials CT/MRI fusion Patients and treatment Patients were treated with either whole pelvis 3D conformal radiotherapy or intensity-modulated external beam radiotherapy using a linear accelerator with 18-MV photons. Concomitant chemotherapy was added using weekly cisplatin 40 mg/m2 of body surface for 5 weeks. Dose was prescribed at the ICRU point and was 45 Gy in 25 fractions (1.8 Gy per fraction). All patients signed a written consent before treatment. The patients underwent MRI scanning before the start of external beam radiotherapy and then again, after finishing external beam radiotherapy before the first and fourth fraction of brachytherapy with applicators in place (GammaMed Fletcher-style
Applicators were used for registration to maximize the accuracy of CT and MRI image fusion within the soft tissue. Primarily, three applicator landmarks were used: the top of the tandem (tandem landmark) and the channels for the interstitial needle located nearest to the rectum (ovoid landmarks). The channels are very well visualized on CT and, because of the gel used, also on MRI. In rare cases of deformed ovoids, or different mutual positions of the tandem and ovoids on CT and MRI, additional landmarks were used. Having registered the match points, the fusion was corrected manually using registration tools of the treatment planning software which enabled shifts and rotation in all planes. All image fusions were performed by the same person.
148
Dolezel et al.
International Journal of Radiation Oncology Biology Physics
3D planning After the registration procedure was completed, we reconstructed the applicators using CT images. The contours of OAR were modified according to current topography on the CT images. Gross tumor volume (GTV), HR CTV, and intermediate-risk (IR) CTV contours were used from the MRI images. Treatment times for dwell positions were transferred from the preplan. Because of the modification of the contours of OAR, adaptation of the treatment plan was always necessary. We followed doseevolume constraints for OAR and for target identical to those used for preplanning.
Registration evaluation The accuracy of CT/MRI registration was evaluated in 6,132 measurements from 42 applications in 7 consecutive patients with CT/MRI data fusion performed in two steps, assessing fusion agreement of both applicators and soft tissue. Every image fusion was monitored by 146 measurements. First, the position of the applicators on CT and MR images was compared using 14 measurements. Absolute positional differences of the center of tandem on CT and MRI was evaluated in four transverse planes (top of the tandem, 15 mm and 30 mm from the top of the tandem, and 10 mm above the external orifice of the cervical canal), in both anteroposterior and laterolateral directions. Differences between the centers of the ovoids on CT and MRI scans were also measured in both directions. Additionally, both the distance between the center of the ovoids and applicator rotation were evaluated. Second, an analysis of registration accuracy of soft tissue was performed. Because of the unequal volume of the cervix on CT and MRI images, differences in the relative position of the applicator within the uterine cervix and body were determined on CT and MRI scans using the ratio of the distance between the applicator and external uterine contour to the uterine diameter in eight angular projections (0 e180 , 45 e225 , 90 e270 , 135 e315 ) and four transverse planes (top of the tandem, 15 mm and 30 mm from the top of the tandem, and 10 mm above the external orifice of the cervical canal). The difference in the relative applicator position on CT and MRI was converted to the absolute distance in millimeters which indicated the difference between the real and ideal applicator position (i.e., agreement of soft-tissue fusion) (Fig. 1aeb). Measurements were performed manually using TPS tools.
Dosimetric evaluation Four plans were compared for dosimetric evaluation. Besides the two plans used for treatment planning (MRI preplan and definitive CT/MRI fusion plan), other two modifications of CT/MRI fusion plans were created to analyze the dosimetric impact of volume variability: CT/MRI fusion plan without OAR adaptation and CT/ MRI fusion plan without target adaptation. CT/MRI fusion plan without OAR adaptation was created after CT/MRI fusion by using dwell times from MRI preplan regardless of the changed topography of organs at risk (rectum, sigmoid, bladder) to assess the need for inter-fraction reimaging. CT/MRI fusion plan without target adaptation was planned by using target volume contours from MRI performed before first brachytherapy fraction; however, doseevolume histogram
Fig. 1. (a) Diagram of fusion analysis (soft tissue). The ratio of the distance between the applicator and uterine margin to uterine diameter was determined on computed tomography (CT) and magnetic resonance imaging (MRI) scans in eight directions, at angular intervals of 45 . Accuracy of registration of soft tissue was evaluated for each fraction on transverse images in 15-mm steps from the top of the tandem to the cervix, using at least 132 measurements. (Example of absolute difference of the relative position of the applicator in direction 90 e270 Z a/b - d/c.) (b) Diagram of fusion accuracy assessment (soft tissue). The absolute difference of the relative position was converted to the absolute distance in millimeters (z). In this example, in direction 90 e270 . parameters for target volumes (GTV, HR CTV, IR CTV) were reported to MRI performed before fourth brachytherapy fraction to determine the dosimetric impact of spatial variability of target volume contours. In addition, the GTV volume regression during radiotherapy was measured (before start of external beam radiotherapy, before fraction 1 of brachytherapy, and before fraction 4 of brachytherapy).
Statistics All continuous numerical variables were presented as mean values with ranges and standard deviations. Differences in position of the
Volume 84 Number 1 2012 Table 1
MRI-based preplanning with CT/MRI fusion in 3D GYN BRT
149
Inaccuracy of CT/MRI applicator registration Laterolateral direction (mm) (mean [SD]) (median [range])
Tandem 1.10 Top of the tandem 1.28 15 mm below top of the tandem 1.16 30 mm below top of the tandem 1.06 10 mm above the external orifice of cervical canal 1.14 Ovoid Center of the left ovoid 1.68 Center of the right ovoid 1.73 Distance between center of the ovoids 2.88 mm Rotation of the applicator 3.32
(0.61) (0.83) (0.52) (0.45) (0.42)
1.10 1.15 1.15 1.05 1.00
(0.30e3.60) (0.60e3.60)* (0.40e2.60)* (0.30e2.60)* (0.3e2.40)*
(0.67) (0.73) (0.86) (1.56)
1.60 (0.50e4.80) 1.70 (0.40e5.10)
Anteroposterior direction (mm) (mean [SD]) (median [range]) 1.20 1.52 1.38 1.08 0.98 1.87 1.96 2.85 mm 3.10
(0.68) (0.82) (0.56) (0.41) (0.34)
1.10 1.50 1.25 1.10 1.00
(0.20e3.40) (0.40e3.40)y (0.30e2.90)y (0.20e2.10)y (0.30e1.90)y
(0.72) 1.80 (0.50e3.60) (0.65) 1.90 (0.40e3.80) (0.80e5.10) (1.00e7.40)
Abbreviation: SD Z standard deviation. * The differences between measured values were not significant. y The differences between measured values were significant between planes with a distance of 30 mm or more (top of the tandem vs. 30 mm below top of the tandem or 10 mm above the external orifice of cervical canal, and 15 mm below top of the tandem vs. 10 mm above the external orifice of cervical canal).
applicators and doseevolume histogram parameters were assessed by t-test. A p value of 0.05 or less was considered significant.
Results The insertion of applicators under paracervical anesthesia in the radiology department was successful in all patients. We observed no complications during the entire MRI procedure at the radiology department (insertion; MRI scanning; applicator removal and subsequent transport). The average time of insertion of applicators in the MRI room under paracervical anesthesia was 15 min. The average contouring and planning time needed for preplanning was 77 min (range, 55e94 min). The average time needed for CT/MRI data fusion, modification of OAR contours, and plan adaptation was 34 min (range, 25e49 min). When assessing the accuracy of applicator registration, the median absolute difference of the center of the tandem on CT and MRI was 1.1 mm in both the anteroposterior and laterolateral directions. There was no statistical difference in the laterolateral direction. However, a statistical difference in the anteroposterior direction between planes of 30 mm or more distance from each other was observed. In more than 85% of measurements, the absolute difference of the tandem on CT and MRI was less than 1.5 mm. In only 7.5% of measurements, the difference was greater than 2 mm. The median distance between the center of the right ovoid on CT and MRI was 1.7 mm and 1.9 mm in the laterolateral and anteroposterior directions, respectively. Similarly, the median distance between the center of the left ovoid was 1.6 mm and 1.8 mm, respectively. More than 93% of measurements had a difference in the distance between right or left ovoid smaller than 2.5 mm. The median absolute difference between the separation of the center of ovoids on CT and MRI was 2.85 mm. The rotation of applicators was 3.1 (Table 1). In terms of soft-tissue registration accuracy, the median absolute difference in the position of the applicator in relation to soft tissue was reduced from 1.93 mm in the transverse plane passing through the top of the tandem to 0.84 mm in the plane 10 mm above the external orifice of the cervical canal. The differences were statistically significant. Similarly, the median absolute difference in the applicator position in relation to soft tissue was
between 1.17 mm and 1.28 mm in selected directions (angles 0 e180 , 45 e225 , 90 e270 , 135 e315 ), without statistical significance (Table 2, Fig. 2). In 80% of measurements, the difference was smaller than 2 mm. An inaccuracy greater than 3 mm was found in only 7.1% of measurements. The degree of inaccuracy of applicator/applicator fusion correlated with the inaccuracy of soft-tissue registration. Three of 42 evaluated fusions were found unacceptable, with an absolute difference of the tandem on CT and MRI of more than 2.0 mm, an absolute difference of ovoid on CT and MRI of more than 2.5 mm, and an inaccuracy of soft tissue greater than 3.0 mm. All incorrect fusions were observed in a single patient. The mean volumes of GTV before treatment, before fraction 1 of brachytherapy, and before fraction 4 of brachytherapy were 65.4 cm3, 12.6 cm3, and 7.8 cm3, respectively. The mean dose
Table 2
Inaccuracy of CT/MRI soft-tissue registration Absolute difference in position of the applicator in relation to soft tissue (mm)
Plane Top of the tandem 15 mm below top of the tandem 30 mm below top of the tandem 10 mm above the external orifice of cervical canal Angle 0 and 180 45 and 225 90 and 270 135 and 315
mean (SD)
median (range)
2.05 (1.48) 1.58 (0.89)
1.93 (0.11e5.87)y 1.50 (0.14e4.23)y
1.12 (0.85)
1.05 (0.13e4.15)y
0.90 (0.64)
0.84 (0.06e3.41)y
1.25 1.37 1.34 1.23
1.17 1.28 1.27 1.17
(1.14) (1.26) (0.88) (0.93)
(0.06e5.64)* (0.11e5.47)* (0.08e4.33)* (0.08e5.87)*
Abbreviation: SD Z standard deviation. * The differences between measured values were not significant. y The differences between measured values were significant.
150
Dolezel et al.
Fig. 2. Distribution of computed tomography/magnetic resonance imaging registration inaccuracy. Each point represents the difference in the position of the applicator in relation to soft tissue: 42 fractions, 4 planes. > Z top of the tandem; O Z 15 mm below top of the tandem; , Z 30 mm below top of the tandem; Δ Z 10 mm above the external orifice of cervical canal.
values to GTV, HR CTV, IR CTV, urinary bladder, sigmoid, and rectum for all four plans are summarized in Table 3. No statistically significant difference in dosimetric parameters between CT/ MRI plans in comparison to MRI preplans was observed; however, dwell times varied significantly. The dosimetric parameters for organs at risk on CT/MRI fusion plans without OAR adaptation were significantly worsened whereas the target coverage was not changed. Planning without target adaptation led to overdosing of the target volume, especially HR CTV.
Discussion Conventional dose calculation for intracavitary brachytherapy was for decades based on the information obtained from two orthogonal radiographs with limited visualization of the organs at risk, and with no information about the tumor itself. Although CT imaging is sufficient for definition of OAR (rectum, bladder, sigmoid), it does not provide sufficient information about the tumor topography nor the distinction between the uterine cervix and corpus. Recently, 3D MRI-guided brachytherapy has been established in clinical practice (2e5, 10, 12e20). The first published clinical results demonstrated a reduction in morbidity and improved local control (13, 14). Because of the absence of an MRI scanner in our radiotherapy department, we aimed to develop an alternative technique of MRI evaluation with applicators in situ using MRI-based preplanning with consecutive CT/MRI data fusion. The accuracy of CT/MRI fusion based on the applicators is highly dependent on the congruent geometry of applicators both at MRI preplanning and CT application. Therefore, we strictly adhered to an identical topography of applicators. We also developed homemade CT/MRI ovoids, compatible table, and homemade MRI-compatible clamping device. As a result, fusion of CT and MRI images with only a 1.1 mm absolute difference in the position
International Journal of Radiation Oncology Biology Physics of tandem and 1.7, respectively 1.8 mm in the position of ovoids has been achieved. Determination of the accuracy of soft-tissue registration is a “sophisticated approximation” because of the unequal volumes of the uterine cervix and body on CT and MRI images. In our opinion, conversion of the absolute difference of the relative position of the applicator in relation to the outer contour of the uterine cervix or body is rather controversial but still the most precise way of evaluation. To maximize the accuracy of our method, we performed a large number of measurements for every CT/MRI fusion (146 measurements per CT/MRI registration). The absolute difference in the position of the applicator in relation to the soft tissue was decreasing towards the cervix (1.93 mm, 1.50 mm, 1.05 mm, 0.84 mm), the differences reached statistical significance. To the contrary, there was no statistical difference in any of the eight measured angular projections. The results are in keeping with our presumption. Because of decreased fixation of the upper part of the uterus in comparison to the cervix, soft-tissue registration based on the applicators showed increasing accuracy toward the cervix. Because the tumor is located in an area with the most precise fusion, we can postulate that this approach is acceptable for image registration. We found unacceptable results of image fusion in three out of 42 evaluated fusions. Interestingly enough, all inappropriate fusions were obtained in the same patient. We observed, in this particular case, significant topographic changes (narrowing) in the upper part of vagina after the first fraction of brachytherapy and a small but significant difference in rotation of the tandem on MRI before the first fraction of brachytherapy compared to the CT scan performed during the first three applications. Because applicators and critical organs are reconstructed and contoured directly on the CT scan, the only, yet very important inaccuracy, applies to the target volumes (GTV, HR CTV, and IR CTV). This fusion inaccuracy of soft tissues must be taken into account during the planning process. We found significant GTV volume regression during external beam chemoradiotherapy and only slight changes of tumor volume during brachytherapy. Treatment times for individual dwell positions differed significantly between preplans and all CT/MRI plans mostly because of the different interfractional topography of OAR. The most variable organ was sigmoid. Minor changes were found for rectum and (likely from routine instillation of 50 mL saline) minimal changes for bladder topography. Therefore, we found the greatest interfractional difference in particular dwell position at the upper part of the tandem (39%e287%), lesser in ovoids (59%e 195%), and minimal in the lower part of the tandem (74%e137%). Due to the substantial interfractional OAR volume changes in comparison with GTV regression during brachytherapy, no significant correlation between the need for OAR adaptation and tumor regression was found. Our dosimetric analysis confirmed the necessity of preplan adaptation and the need for interfraction imaging. Our results show that CT acquisition for every fraction from OAR variability is essential and, in addition, at least two to three MRI acquisitions are advisable because of systematic shrinkage of the HR CTV (Fig. 3, Table 3). The main advantages of MRI-based preplanning with satisfactory CT/MRI data fusion are: precise definition of GTV and HR CTV; increased conspicuity of applicators on CT scans with minimal risk of inaccuracy of applicator reconstruction; decreased time needed for the planning process with a reduction of volume changes of OAR; time-unlimited preplanning (without personal stress and decreased comfort to the patient because of prolonged
Volume 84 Number 1 2012 Table 3
MRI-based preplanning with CT/MRI fusion in 3D GYN BRT
151
Difference in dosimetry between MRI preplan and CT/MRI fusion plans
GTV D100 (Gy a/b10) D90 (Gy a/b10) HR-CTV D100 (Gy a/b10) D90 (Gy a/b10) IR-CTV D100 (Gy a/b10) D90 (Gy a/b10) Bladder ICRU point D0.1cc (Gy a/b3) D1cc (Gy a/b3) D2cc (Gy a/b3) Rectum ICRU point D0.1cc (Gy a/b3) D1cc (Gy a/b3) D2cc (Gy a/b3) Sigmoid D0.1cc (Gy a/b3) D1cc (Gy a/b3) D2cc (Gy a/b3)
MRI preplan (mean SD)
CT/MRI fusion plan without OAR adaptation (mean SD)
CT/MRI fusion plan (mean SD)
CT/MRI fusion plan without target adaptation (mean SD)
93.9 21.2 122.1 28.4
93.7 22.0; *p Z 0.49 122.3 28.6; *p Z 0.61
93.1 22.0; zp Z 0.62 121.3 29.6; zp Z 0.68
95.9 21.3; yp Z 0.10 123.4 29.6; yp Z 0.08
66.1 7.8 83.8 12.7
66.3 7.7; *p Z 0.54 83.9 12.9; *p Z 0.47
65.4 7.6; zp Z 0.59 83.1 13.8; zp Z 0.51
68.5 8.3; yp < 0.05 88.2 11.6; yp < 0.05
53.1 7.4 64.2 7.3
53.2 7.5; *p Z 0.42 64.2 7.3; *p Z 0.39
52.6 7.7; zp Z 0.47 63.8 6.9; zp Z 0.37
54.1 8.2; yp Z 0.13 65.2 7.6; yp Z 0.09
65.5 96.2 82.8 77.5
7.9 6.4 6.1 5.1
66.9 103.4 88.2 84.1
8.8; 9.7; 7.5; 7.3;
*p *p *p *p
Z 0.21 < 0.05 < 0.05 < 0.05
65.7 97.1 83.6 78.2
8.1; 6.7; 6.5; 5.8;
67.2 73.8 67.1 63.6
2.9 3.8 2.7 2.4
72.9 82.3 71.4 69.3
3.4; 3.9; 3.1; 2.4;
*p *p *p *p
< < < <
67.9 74.9 67.5 64.2
2.7; 3.6; 2.3; 1.9;
76.8 4.1 68.7 3.1 64.2 2.7
0.05 0.05 0.05 0.05
90.9 4.5; *p < 0.05 78.7 3.6; *p < 0.05 74.2 3.7; *p < 0.05
z
p p z p z p
Z Z Z Z
0.82 0.54 0.31 0.46
66.1 97.3 84.2 78.6
8.5; 6.9; 6.7; 6.1;
z
Z Z Z Z
0.73 0.62 0.59 0.53
68.3 75.4 67.9 64.6
3.1; 3.9; 2.4; 2.1;
z
p p z p z p z
77.5 3.9; zp Z 0.63 69.4 2.8; zp Z 0.41 65.5 2.9 zp Z 0.38
y
p p y p y p
Z Z Z Z
0.43 0.76 0.31 0.49
y
Z Z Z Z
0.87 0.74 0.79 0.67
y
p p y p y p y
77.8 4.2; yp Z 0.92 69.7 3.0; yp Z 0.75 65.9 3.2; yp Z 0.58
Abbreviations: CT Z computed tomography; D90 Z dose received by at least 90% of volume; D0.1cc, D1cc, D2cc Z minimal dose for most irradiated 0.1 cm3, 1 cm3, and 2 cm3; D100 Z minimal target volume; GTV Z gross tumor volume; HR-CTV Z high-risk clinical target volume; ICRU Z International Commission on Radiation Units and Measurements; IR-CTV Z intermediate-risk clinical target volume; MRI Z magnetic resonance imaging; OAR Z organs at risk; SD Z standard deviation. * Comparison of dosimetric parameters between CT/MRI fusion plans without OAR adaptation and CT/MRI fusion plans (t-test). y Comparison of dosimetric parameters between CT/MRI fusion plans without target adaptation and CT/MRI fusion plans (t-test). z Comparison of dosimetric parameters between CT/MRI fusion plans and MRI preplans (t-test).
contouring and planning process); and the possibility of combined interstitial and intracavitary brachytherapy in patients with a suboptimal response to the external chemoradiotherapy. The disadvantages, obviously, are the additional insertion of applicators at the radiology department under paracervical anesthesia before the first and fourth fraction of brachytherapy without radiotherapy treatment, and the necessity of congruent alignment
of applicators on MRI and CT scans. Also, it would be more appropriate to use a deformable algorithm for accurate coregistration of MRI onto CT.
Conclusions This is the first evidence of the systematic integration of MRI based preplanning with consecutive CT/MRI data fusion into clinical practice. Our results suggest that this technique is feasible and safe with an acceptable inaccuracy of soft-tissue registration. Fusion inaccuracy must be taken into account during the planning process.
References
Fig. 3. Mean absolute tumor volumes during radiotherapy. EBRT Z External beam radiotherapy; BRTZ brachytherapy.
1. Po¨tter R, Dimopoulos J, Bachtiary B, et al. 3D-conformal HDR brachytherapy and external beam therapy plus simultaneous cisplatin for high-risk cervical cancer: Clinical experience with 3 year follow up. Radiother Oncol 2006;79:80e86. 2. Po¨tter R, Fidarova E, Kirisits C, et al. Image-guided adaptive brachytherapy for cervix carcinoma. Clin Oncol 2008;20:426e432. 3. Dimopoulos JCA, Schard G, Berger D, et al. Systematic evaluation of MRI findings in different stages of treatment of cervical cancer: Potential of MRI on delineation of target, pathoanatomic structures, and organs at risk. Int J Radiat Oncol Biol Phys 2006;64:1380e1388.
152
Dolezel et al.
4. Dimopoulos JC, Schirl G, Baldinger A, et al. MRI assessment of cervical cancer for adaptive radiotherapy. Strahlenther Onkol 2009; 185:282e287. 5. Haie-Meder C, Chargari C, Rey A, et al. DVH parameters and outcome for patients with early-stage cervical cancer treated with preoperative MRI-based low dose rate brachytherapy followed by surgery. Radiother Oncol 2009;93:316e321. 6. Haie-Meder C, Po¨tter R, van Limbergen E, et al. Recommendations from the Gynaecological (GYN) GEC ESTRO Working Group: Concepts and terms in 3D-image based 3D-treatment planning in cervix cancer brachytherapy with emphasis on MRI assessment of GTV and CTV. Radiother Oncol 2004;74:235e245. 7. Po¨tter R, Haie-Meder C, vanLimbergen E, et al. Recommendations from gynaecological (GYN) GEC ESTRO working group (II): Concepts and terms in 3D image-based treatment planning in cervix cancer brachytherapy-3D dose-volume parameters and aspects of 3D-image-based anatomy, radiation physics, radiobiology. Radiother Oncol 2006;78:67e77. 8. Po¨tter R, Dimopoulos J, Georg P, et al. Clinical impact of MRI assisted dose volume adaptation and dose escalation in brachytherapy of locally advanced cervix cancer. Radiother Oncol 2007;83:148e155. 9. Dimopoulos JCA, Lang S, Kirisits C, et al. Doseevolume histogram parameters and local tumor control in MR image guided cervical cancer brachytherapy. Int J Radiat Oncol Biol Phys 2009;75:56e63. 10. Kirisits C, Po¨tter R, Lang S, et al. Dose and volume parameters for MRI based treatment planning in intracavitary brachytherapy of cervix cancer. Int J Radiat Oncol Biol Phys 2005;62:901e911. 11. Lang S, Kirisits C, Dimopoulos J, et al. Treatment planning for MRI assisted brachytherapy of gynaecological malignancies based on total dose constraints. Int J Radiat Oncol Biol Phys 2007;69:619e627. 12. Haack S, Nielsen SK, Lindegaard JC, et al. Applicator reconstruction in MRI 3D image-based dose planning of brachytherapy for cervical cancer. Radiother Oncol 2009;91:187e193.
International Journal of Radiation Oncology Biology Physics 13. Tanderup K, Hellebust TP, Lang S, et al. Consequences of random and systematic reconstruction uncertainties in 3D image based brachytherapy in cervical cancer. Radiother Oncol 2008;89: 156e163. 14. Muschitz S, Petrow P, Briot E, et al. Correlation between the treated volume, the GTV and the CTV at the time of brachytherapy and the histopathologic findings in 33 patients with operable cervix carcinoma. Radiother Oncol 2004;73:187e194. 15. Kirisits C, Lang S, Dimopoulos J, et al. The Vienna applicator for combined intracavitary and interstitial brachytherapy of cervical cancer: Design, application, treatment planning and dosimetric results. Int J Radiat Oncol Biol Phys 2006;65:624e630. 16. Dimopoulos JCA, Kirisits C, Petric P, et al. The Vienna applicator for combined intracavitary and interstitial brachytherapy of cervical cancer: Aspects of clinical feasibility and preliminary results. Int J Radiat Oncol Biol Phys 2006;66:83e90. 17. Van de Kamer JB, De Leeuw AA, Moerland MA, et al. Determining DVH parameters for combined external beam and brachytherapy treatment: 3D biological dose adding for patients with cervical cancer. Radiother Oncol 2010;94:248e253. 18. Ju¨rgenliemk-Schulz IM, Lang S, Tanderup K, et al. Variation of treatment planning parameters (D90 HR-CTV, D 2cc for OAR) for cervical cancer tandem ring brachytherapy in a multicentre setting: Comparison of standard planning and 3D image guided optimisation based on a joint protocol for dose-volume constraints. Radiother Oncol 2010;94:339e345. 19. Beriwal S, Kim H, Coon D, et al. 3D HDR intracavitary brachytherapy for the treatment of cervical cancer: Feasibility and early results from single institution. Int J Radiat Oncol Biol Phys 2009; 75:367e368. 20. Holloway CL, Racine ML, Cormack RA, et al. Sigmoid dose using 3D imaging in cervical-cancer brachytherapy. Radiother Oncol 2009;93: 307e310.